Genetic prediction models, poised to revolutionize medicine, are currently hampered by a significant flaw: they are predominantly trained on DNA from individuals of European ancestry. This reliance on a narrow genetic dataset means these powerful tools do not work equally well for everyone.

Consequently, these models may not accurately assess health risks for people from diverse genetic backgrounds, potentially leading to misdiagnoses or missed diagnoses. This could exacerbate existing health care disparities, offering advanced predictive capabilities to some populations while leaving others behind.

As these technologies become more integrated into healthcare, it is crucial to address the inherent biases in their training data. Ensuring more inclusive and representative datasets is essential for genetic risk tools to fulfill their promise of improving health outcomes for all individuals, regardless of their ancestry.